spacex-skill

SpaceX: an Agent Skill for weird ideas you can act on

A two-phase ideation skill for Claude. Liftoff thinks about a real problem the way a stoned person thinks about a doorknob: sincere, associative, every connection feeling cosmic. Reentry is mission control reading the flight recorder the next morning: it keeps only ideas that are both sound and absent from a sealed "sober baseline," then turns them into moves you can test this week.

The output is a mission report with at most three ideas, each with a concrete move and a cheapest test, plus telemetry on how many claims survived.

Simulation only. The skill copies how associative thinking works; it never mentions or discusses substances.

Install

Claude Code (plugin marketplace)

/plugin marketplace add cumeadi/spacex-skill
/plugin install spacex@spacex-skill

Claude Code (manual)

Copy skills/spacex into ~/.claude/skills/ (personal) or .claude/skills/ in a project.

Claude.ai / Claude desktop

Download spacex.zip from the latest release, then upload it under Settings → Capabilities → Skills.

Claude API

Upload the skills/spacex folder through the Skills API. See Anthropic's Skills docs.

Use

Say "SpaceX" or "get high on this problem" and give it something real:

SpaceX: meal-kit subscribers cancel after week six. Dose 2.

• Dose 1 (low orbit), 2 (orbit, default), 3 (deep space)
• Register giggly, cosmic (default), or paranoid-philosophical

Layout

skills/spacex/
├── SKILL.md              # instructions and frontmatter
└── scripts/
    └── roll_moves.py     # rolls each trip's move order

License

MIT. See LICENSE.